From mboxrd@z Thu Jan  1 00:00:00 1970
From: Andy Wingo <wingo@igalia.com>
Subject: bug#28211: Stack marking issue in multi-threaded code, 2020 edition
Date: Tue, 17 Mar 2020 22:16:22 +0100
Message-ID: <87a74eznq1.fsf@pobox.com>
References: <877exuj58y.fsf@gnu.org> <87d0yo1tie.fsf@gnu.org>
 <87fu3124nt.fsf@gnu.org> <87d0y5k6sl.fsf@netris.org>
 <871sel6vnq.fsf@igalia.com> <87fu30dmx3.fsf@netris.org>
 <87tvrg3q1d.fsf@igalia.com> <87a7rdvdm9.fsf_-_@gnu.org>
 <87tv2tp74g.fsf_-_@gnu.org>
Mime-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable
Return-path: <bug-guix-bounces+gcggb-bug-guix=m.gmane-mx.org@gnu.org>
Received: from eggs.gnu.org ([2001:470:142:3::10]:38780)
 by lists.gnu.org with esmtp (Exim 4.90_1)
 (envelope-from <Debian-debbugs@debbugs.gnu.org>) id 1jEJaC-0007Ym-S6
 for bug-guix@gnu.org; Tue, 17 Mar 2020 17:17:06 -0400
Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71)
 (envelope-from <Debian-debbugs@debbugs.gnu.org>) id 1jEJaA-00082S-HO
 for bug-guix@gnu.org; Tue, 17 Mar 2020 17:17:04 -0400
Received: from debbugs.gnu.org ([209.51.188.43]:33638)
 by eggs.gnu.org with esmtps (TLS1.0:RSA_AES_128_CBC_SHA1:16)
 (Exim 4.71) (envelope-from <Debian-debbugs@debbugs.gnu.org>)
 id 1jEJaA-0007zo-Bu
 for bug-guix@gnu.org; Tue, 17 Mar 2020 17:17:02 -0400
In-Reply-To: <877exuj58y.fsf@gnu.org>
Sender: "Debbugs-submit" <debbugs-submit-bounces@debbugs.gnu.org>
Resent-Message-ID: <handler.28211.B28211.1584479804536@debbugs.gnu.org>
List-Id: Bug reports for GNU Guix <bug-guix.gnu.org>
List-Unsubscribe: <https://lists.gnu.org/mailman/options/bug-guix>,
 <mailto:bug-guix-request@gnu.org?subject=unsubscribe>
List-Archive: <https://lists.gnu.org/archive/html/bug-guix>
List-Post: <mailto:bug-guix@gnu.org>
List-Help: <mailto:bug-guix-request@gnu.org?subject=help>
List-Subscribe: <https://lists.gnu.org/mailman/listinfo/bug-guix>,
 <mailto:bug-guix-request@gnu.org?subject=subscribe>
Errors-To: bug-guix-bounces+gcggb-bug-guix=m.gmane-mx.org@gnu.org
Sender: "bug-Guix" <bug-guix-bounces+gcggb-bug-guix=m.gmane-mx.org@gnu.org>
To: Ludovic =?UTF-8?Q?Court=C3=A8s?= <ludo@gnu.org>
Cc: 28211@debbugs.gnu.org

On Thu 12 Mar 2020 22:59, Ludovic Court=C3=A8s <ludo@gnu.org> writes:

> I think I=E2=80=99ve found another race condition involving stack marking=
, as a
> followup to <https://issues.guix.gnu.org/issue/28211> (this time on
> 3.0.1+, but the code is almost the same.)
>
> =E2=80=98abort_to_prompt=E2=80=99 does this:
>
>   fp =3D vp->stack_top - fp_offset;
>   sp =3D vp->stack_top - sp_offset;
>
>   /* Continuation gets nargs+1 values: the one more is for the cont.  */
>   sp =3D sp - nargs - 1;
>
>   /* Shuffle abort arguments down to the prompt continuation.  We have
>      to be jumping to an older part of the stack.  */
>   if (sp < vp->sp)
>     abort ();
>   sp[nargs].as_scm =3D cont;
>   while (nargs--)
>     sp[nargs] =3D vp->sp[nargs];
>
>   /* Restore VM regs */
>   vp->fp =3D fp;
>   vp->sp =3D sp;
>   vp->ip =3D vra;
>
>
> What if =E2=80=98scm_i_vm_mark_stack=E2=80=99 walks the stack right befor=
e the =E2=80=98vp->fp=E2=80=99
> assignment?  It can determine that one of the just-assigned =E2=80=98sp[n=
args]=E2=80=99
> is a dead slot, and thus set it to SCM_UNSPECIFIED.

I think you're right here.

Given that the most-recently-pushed frame is marked conservatively, I
think it would be sufficient to reset vp->fp before shuffling stack
args; that would make it so that the frame includes the values to
shuffle, their target locations, and probably some other crap in
between.  Given that marking the crap is harmless, I think that would be
enough.  WDYT?

In a more perfect world, initiating GC should tell threads to reach a
safepoint and mark their own stacks -- preserves thread locality and
prevents this class of bug.  But given that libgc uses signals to stop
threads, we have to be less precise.

Cheers,

Andy